Berkeley, CA
- Tweets 17,391
- Following 1,616
- Followers 23,768
- Likes 22,290
As aesthetic fodder I love looking at these things. But I don't understand how it really helps anyone make sense of their codebase.
I've tried like 10+ abstractions of code I've been building and really struggle at finding any "high altitude" view of the work being done that is better than reading a doc. But perhaps I'm missing something others find great.
system-atlas: a skill that builds an explorable isometric map of your codebase / agent / pipeline from one data file
arranging the system visually does something for my brain that a wall of text doesn't
i've found it way more useful to build up understanding of a system or a large change through these progressive disclosure style atlases than through docs.
hover to read, click to pin, step inside a box to watch it run, click a packet to see the actual payload. chapters reveal the system a few boxes at a time instead of all at once. the md twin is generated from the same file so it can't drift
works from both sides.
- building a new architecture and moving pieces around visually before anything is written
- pointed it at a 67 pr stack (don't ask how i got there) and finally understood how the thing fit together
it also keeps track of every open question in the system. as you clarify or improve things the answers land in the atlas, so "wait did we decide that" has an answer instead of a scroll through slack.
1
19
3,467
Its 2026 and I'm trying to buy a movie ticket on the Internet.
They should really point Claude at this thing and tell it to fix it.
2
11
1,379
Everybody likes making.
It's the maintenance that gets ya.
I think this "everyone will build" makes sense for software that is mostly a one-time build or low maintenance product. But I have a lot of mothballed projects now that are a pain to maintain.
For things that are bleeding edge and move constantly, hard to imagine folks don't want to "hire" the best product (and co behind it) to do that for them unless it is truly core to them. Cloud agent orchestration? Very hard to stay on ahead.
1
16
1,686
like you have to want to do stuff like this every week for years and assume you are in it better than the folks who it's their main job vs a fun side project while you are trying to world class at your thing.
conductor.build/changelog
5
495
Built a mac app this week using the new Conductor API that just came out. If you use @conductor_build you may like it.
It's called Whistle, and it's for those quick ideas you have on a project.
It's a tray app, you click it, you start talking or typing an idea that just came to mind you want Conductor to work on for a bit. Then move on with your day, next time you open up Conductor it's done the research, prepped a doc, and you are ready to go.
Hope you like.
nabeelhyatt.com/experiments/…
6
1
39
16,037
Mostly made this for myself because I'm often running out of a meeting, or in mid email or something, and have an idea. Opening big boy Conductor can feel distracting ("ooh it got back to me on the thing") so I just wanted a way to stay in flow, throw an idea out there and have it research.
It's using apple native voice reco for now, nothing on my servers, and just pushes to conductor api with a template of instructions on what to do that you can edit.
1
714
Cloud is great, enabling multiplayer is better, enabling API calls into the agent from other workspaces, other colleagues, other agents, has felt magical.
Introducing Conductor Cloud!
Out with worktrees, in with multiplayer cloud workspaces.
Bring your subscriptions, invite your teammates, start your agents, and shut your laptop. Conduct from iPhone and API, too.
conductor.build
1
1
14
2,628
Man, the @meetgranola team has some kind of hall of fame ability to stare at the thing you've always stared at, and ship the obviously intuitive answer that no one did.
The best wearable is... the one you're already wearing.
Excited to announce Granola for Apple Watch. Launching today⌚️
2
2
29
3,472
Strange arc for one word:
"AI" used to culturally stand for "optimized to a superhuman extent" and now means "the fast but sloppy version of a thing"
5
4
47
4,009
Unlike in creative writing, I find the repetitive rhythms of AI writing for coding are a net benefit. But also a subtle lock in.
My ability to work quickly in flow with AI is directly related to my ability to skim the text responses from the AI. Get the gist and fire back a message (who's reading all that). You stay in flow when you know what to expect.
But that also means switching from Claude to some other model, for me, breaks flow substatially. It's not whether the model "does good work" or not, it's that the language affordances are what I'm attuned to.
Switching from Claude to something else is no different than moving from one spreadsheet tool to some new novel tool where all the buttons are in the wrong place. It's just that, with language. It instantly slower, I have to slow my reading pace, figure out where the important bits are. Etc.
I had not really anticipated that and kinda interesting.
1
9
2,616
Will miss you @om
I loved Om's frankness. He just told it like it is, without either kid gloves or venom, a rare trait today in tech. This is my favorite photo of him, from a walk around Lands End, because you can tell in this moment he's lovingly dropping a truth bomb. Great man.
1
4
98
3,520
Asking Fable to "read the relevant academic literature on this idea before building - then think adversarially" is my favorite thing in the last 48 hours.
2
35
3,517
This is actually a really good example of why vertical AI is so important.
Fable spent huge tokens to transcribe, pick a couple takes, and merge them. But @descript would have done that in the first 5 seconds (using Claude) and then in addition it would have given Fable the super powers to spend it's time removng the uhs and ums from the better cut, fix the sound, color grade, fix eye contact, and more.
It puts Fable in a better car vs trying to build the car from scratch. That's ostensibly the role for most companies now.
Lots of people asked how I used Fable to edit its own launch video so I made a video about that!
TLDR it wrote a lot of code & tool calls to use transcription services, ffmpeg, do colorgrading, use the figma mcp, make remotion UI and render it.
I didn't touch a video editor.
7
2
133
41,636
🙌
Most VCs shied away from Anthropic in 2023. Not Yasmin Razavi.
The Spark Capital partner led a $450M round when Anthropic had no public product, no revenue and a massive capital need. Now the AI giant’s rise has landed her on the #ForbesMidas List for the first time. forbes.com/sites/iainmartin/…
📸: Guerin Blask For Forbes
36
6,497
Nabeel Hyatt retweeted
So fun working with the @vercel team. They ship fast.
Conductor has the distinct edge of being an IDE born for coding agents. An ADE if you will.
Agents will take remote dev mainstream. At most large and sensitive organizations, “local” isn’t a thing. But this future wasn’t evenly distributed.
Excited to power their Sandbox!
10
1
82
10,456
Building all your own software is still amazing.
Maintaining it turns out to be less joyful.
Strangely, I think I'd be much more likely to pay a monthly fee for someone to help maintain software I built, than to pay for some company to maintain their software for me.
I suspect others feel similar, there's an unlock in that.
(and no, it's not to maintain the stuff that AI can just auto-pr / auto-eval)
5
13
1,372